Except that computers should easily be programmed to take into account splits at different positions.
2009 Rangers (Tex/Cano/Jeter/A-Rod) The Dodgers in the 1970's The 1982 (?) Brewers get my vote (Cooper/Gantner/Yount/Molitor). Yount was an absolute MONSTER with an incredible WAR of 11.
I would imagine the assumption is that a hitter should hit similarly regardless of position he plays at. So in terms of predictive value, it probably assumes no difference in him as a hitter as a C and a DH. Obviously, there are anomolies and Gattis appears to be one of those, but I'm not sure their model would want to take that into account if they believe it's mostly randomness.
It just seems the model would produce results for each player’s offensive output at any given position. It would take me less than a minute to find out a player’s splits since that data exists and is easy to find. So, some computer spits it out.
Sure - but you're assuming the model *wants* to account for hitters hitting differently when they play different positions. If the designers of the model think that's just statistical noise, it would want to ignore that information.
Computer doesn't know what position a player will play, but is making a projection based on what a player has done compared to model of based on historical players. The writer is the one speculating that his offensive numbers may improve if he doesn't catch.
That's just bad/incomplete programming and/or analysis. For example, Marwin is an extremely mediocre offensive player at SS over the past 3 seasons while he is good to very good at the other positions he has played. Their model is poorly designed if they don't take position play into account. (I am not arguing with you...but with the design of the analysis if that is the case)
The Rangers? Not sure how you mix up the Rangers and the Yankees. Astros could certainly be the best. I think the Brewers look extra impressive if you thrown in catcher (so would the Yankees with Posada). Ted Simmons was an 8 time all-star.
I mixed up the Rangers and Yankees probably because I was thinking that Tex and A-Rod were on the Rangers together. Yes I meant the Yankees. My vote goes to that Brewers team. They could beat you in so many ways. They had some power, but could run, get on base and were solid with the gloves. Plus Yount was an absolute monster that season.
If they went to a model with manually entered positions instead of projecting the positions, it likely would not affect a player like Marwin as offensive splits based on position are not likely predictive except for C, DH, and PH. Even with someone like Gattis, the model would likely regress the numbers a lot based on sample sizes.
Honestly, if it wasn't for Jeter's name, I wouldn't have even noticed the error. The 2009 Yankees would be a perfect answer with 5 star players, all of whom were still very good players at the time (no young guys before they broke out or old washed up guys), but of course none of those 5 were having MVP type years.
But the question is whether any of that has any predictive value of not. In 2015, Marwin was best as a 2B and 3B. In 2016, he was best as a 1B. In 2017, he was good everywhere except SS. What does that tell us about 2018? You're suggesting it gives us info (that he's more likely to suck at SS). But they could look at larger data over thousands of players and determine it really tells us nothing - that history suggests he's just as likely to be good at SS as anywhere else next year and that these past years are fluky. Who's really right? We won't know until after the season, but if there's data to support their version, then I don't think it's a bad model - it's just a different model than you use. The question is ultimately whether positional data from the past is really predictive. It's the same age-old debate about whether there are really clutch hitters and such (are hitters consistently better in RBI situations or Late & Close, etc).
Another pretty good one and also featured a pretty good catcher... I guess it is safe to say this is not the first ridiculous IF in baseball history. One thing different about ours is that ours is home grown (admittedly somewhat cheating given Gurriel is really Cuban league grown) and we have Marwin Gonzalez as the primary backup for all 4 positions.
The Big Red Machine was the first team I thought of when I made my original post. But Pete Rose didn't move to 3B until 1975, at which point Tony Perez was no longer as great as he was in his prime. 75/76 Perez is very similar to what Yuli was last season. Pete Rose was better than Bregman in those two years, but not by as much as you would think. I think Bregman could be a 5-6 WAR player in 2018. Altuve is amazingly the only position we have a significant disadvantage, because Joe Morgan was an absolute monster his first 5 years in Cincy. Likewise, if Correa maintains 2017 level for a full season, he gives us a significant advantage at SS. Concepcion was all glove.
